mlaretallack

Running 20 agents at the same time means a lot more task switching, its closer to team management, you may have one agent doing basic grunt work, so can be let alone, another one may be planning and designing, so needs a little bit more q&a, and others may be problem solving, and that does take a lot more supervision. But I have never felt like I have any free time, or down time, at the end of the day I feel far more mentily exhausted (in a good way) than any day before agentic development.

A lot of the tools we have to use are designed for meat proxy's (JIRA etc...) yes, you can allow the agents access to the task repo via api or MCP, but at the end of the day, the tool starts to become the bottleneck, Iike the idea of putting the tasks and issues into git, it means all info for why a change was made is in one place, accessable forever, just need to expand agents to non-software dev world.

Did not think I would like that, but really good, simple and easy summary.

The local councils (local Authority) usually get external companies to design the systems, but there are standard, off the shelf solutions that are specified to be used. These systems range from time synced junctions (with Vehicle attenuation) up to full Urban Traffic Control systems, the protocols used are mostly standard (UTMC), so the local Authority just needs to say what they need. For a UTC system it used to be that the council would run the instation, but with cloud computing, that's mostly run for them now as a hosted system. Usually this is all controlled by the standard local council procurement rules, and run by council employees not elected officials. The elected officials can say a scheme is needed and approve costs, but not who does it.

ASCT is standard practice (and has been standard practice) on most junctions in most towns and cities in the UK for over 20 years.

This is a general principal in the eng shop I work for, its written into every step, including the SCRUM etc.. What when well, why did it fail, what could be done better. Its all part of root cause analysis.

I personally use it for the agents I use, so why did it fail, what was the root cause, what can we do to prevent it again.

Very cool, Its interesting to see what happens to Antarctica.

Loads of them in the UK. Especially in villages where it would be too expensive to install a bridge. Just look at the Southampton to Weymouth line.

I hate to ask but what's a "sexy toasters."

Missing Kiro IDE/CLI/Web.

A port of the open epaper lib used in home assistant, but cli based, and an mqtt interface to allow it to run on a different computer to.HA

https://github.com/mretallack/OpenEPaperCliTool

----- 3d printer pipeline, so its can print stuff directly without having to use the computer to set it up.

https://github.com/mretallack/3dprinter

----- Experiment with creating a Abdroid Auto app for phones that cannot run real AA. (WIP)

https://github.com/mretallack/AndroidAuto

----- A android 3d clay modeler to create models for 3d printer, with stl export.

https://github.com/mretallack/ClayModeller

----- Uk Fuel finder python lib and Home Assistant intergration for showing fuel stations from UK gov api.

https://github.com/mretallack/ukfuelfinder https://github.com/mretallack/ukfuelfinder-ha

---- Reverse engineer cheep drone video feed, from drone found in charity shop

https://github.com/mretallack/DroneCamera

---- App to send voice to camera using mqtt.

https://github.com/mretallack/CameraSpeaker

---- Added ONVIF to an oss rtsp android app.

https://github.com/mretallack/cams

---- Added Home Assistant to Dicio Assistant.

https://github.com/mretallack/dicio-android

---- Added telegram bot interface to kiro, with group support.

https://github.com/mretallack/kiro-remote

In Linux, when a shared lib is loaded by multiple processes, its loaded once and not duplicated in ram. Only if a memory page is modified by the process will the memory be duplicated. (Hope I have explained that correctly)

A while ago I created a telegram bridge for AWS Kiro CLI, this allows my to talk to the agent running on my server from anywhere. Any remote access to any of these agents is a massive game changer, it means that you don't need to hover infront of the pc while it works away at the problems. It changes your workflow, but I do find you need to force yourself to "turn off", its easy to do that with the PC, eg, just walk away, but when you can just "get the agent to do one more change" while waiting to pick the kids up or taking the dog for a walk, it can get difficult to stop.

Found out about this today, up until now 802.11p hardware is very expensive, and so you cannot easily do anything with V2x messages like CAM or SPAT, but the fact this was done with sub £20 hardware is really interesting.

100% agree, I find that sometimes I hit a dead end, but the things I build or learn on the way are usable at a later date.

I saw the RAC one this morning, though I was miss reading the graph, as why would the RAC publish such an obvious mistake.

I have written my own Home Assistant custom component for the UK fuel finder data, and yes, the data really is that bad.

I use AWS Kiro, with the Claude models, and its only to happy to help. I give it the headerless ghidra, and decompilers etc... and away it goes.

Yes, I follow the same sort of pattern, it took a while to convince myself that it was ok to leave the agent waiting, but it helps with the human context switching. I also try to stagger the agests, so one may be planning and designing, while another is coding, that way i can spend more time on the planning and designing ones and leave the coding one to get on with it.

Yes, I mostly do spec driven developement. And at the design stage, I always add in tests. I repeat this pattern for any new features or bug fixes, get the agent to write a test (unit, intergration or playwright based), reproduce the issue and then implement the change and retest etc... and retest using all the other tests.

Its very important to understand the "how" it was done. The GPL hands the "compile" step, and the result is still GPL. The clean Room process uses 2 teams, separated by a specification. So you would have to

1. Generate specification on what the system does. 2. Pass to another "clean" system 3. Second clean system implements based just on the specification, without any information on the original.

That 3rd step is the hardest, especially for well known projects.

I use AWS Kiro, and its spec driven developement is exactly this, I find it really works well as it makes me slow down and think about what I want it to do.

Requirements, design, task list, coding.

I also try to avoid negative instructions. No scientific proof, just a feeling the same as you, "do not delete the tmp file" can lead too often to deleting the tmp file.

Not the best way to do it, but I use xfce, multiple workspaces, each with there own version of AWS Kiro, and each kiro has its own project I am working on. This allows me to "switch context" easier between each project to check how the agents are getting on. Kiro also notifies me when an agent wants somthing. Usually I keep it to about 4 projects at a time, just to keep the context switching down.

I agree with this, I put myself in the "glorious hacks to bend the machine into doing things it was never really intended to do" camp, so the end game is somthing cool, now I can do 3 cool things before lunch instead of 3 cool things a year

I am currently doing 6 projects at the same time, where before I would only of doing one at a time. This includes the requirements, design, implementation and testing.

This is where I think its going, it feels that in the end we will end up with an "llm" language, one that is more suited to how an llm works and less human.

The api seems very unstable (or it might just be me) keeps timing out when I try to fetch.

Very interesting, I had a go with Ghidra and AWS Amazon Q, used it to reverse the video feed on a toy drone. I did not think to look for GhidraMCP, would of made it a lot quicker.

I use moto devices, my current one is a g45. But I have also setup a second hand g30.